. Direct Numerical Simulation

What of the future of CFD? Most flows of practical interest are turbu­lent flows. Turbulence is still one of the few unsolved problems in clas­sical physics. In the calculation of turbulent flows, we therefore have to model the effect of turbulence. Any turbulence model involves some empirical data, and all models are inaccurate to some greater or lesser degree. The uncertainty in turbulence models is the reason for much uncertainty in the calculation of turbulent flows in computational fluid dynamics. This will continue for years to come. There is, however, an approach that requires no turbulence modeling. Nature creates a tur­bulent flow using the same fundamental principles that are embodied in the Navier-Stokes equations. Indeed, turbulence on its most detailed scale is simply a flow field developed by nature. If one can put enough grid points in the flow, then a Navier-Stokes solution will calculate all the detailed turbulence without the need for any type of model. This is called direct numerical simulation. The key is "enough grid points,” which even for the simplest flow over a flat plate requires millions of points.

Once again, NASA researchers have been leading the way. Calculations made at NASA Ames for flow over a flat plate have required over 10 mil­lion grid points taking hundreds of hours on supercomputers, an indica­tion of what would be required to calculate the whole flow field around a complete airplane using direct numerical simulation. But this is the future, perhaps, indeed, as far as three decades away. By that time, the computational power of computers will have undoubtedly continued to increase many-fold, and, as well, NASA will be continuing to play a leading role in advancing CFD, even as it is today and has in the past.